Loading ClickForArt…

Blur Text

Blur text in a screenshot or photo: a box of heavy blur you move and size over words, names or numbers. Save it as PNG, JPG or WebP; nothing is uploaded.

The whale photo with a box of heavy blur across the whale's middle and everything else left sharp

What is a text blur?

A text blur softens part of an image until you can't read the writing in it, like an email address in a screenshot or a name on a badge. Blurred words still show that something was there, so the screenshot keeps its layout and only loses the detail.

How to blur text in a screenshot

Load a screenshot or a photo. You'll see a box of the Heavy blur preset across the middle, half the frame wide and about a fifth as tall. Open Mix under Blur: Centre X and Centre Y move the box, Width and Height resize it, and Feather softens the edges.

Radius is how strong the blur is, from 0 to 20. Heavy is 15, Soft is 5 and Light is 2. Tick Invert to blur everything outside the box, or set Mask to Ellipse for a rounder patch.

The page handles one box at a time. For a second patch, save the PNG, load it back in and move the box to the next spot.

How strong does a blur need to be to hide text?

That depends on how tall the type is compared to the picture, because Radius counts pixels of a 1080-pixel-tall frame and scales with the picture. We set [email protected] in 16-pixel Arial, dark on white like a line from a light-mode screenshot, on a 1080-pixel-tall frame and ran it through this blur's own two passes. At Radius 2 (Light) it was still easy to read, and at 5 (Soft) you could make it out. At 10 all that was left were the word shapes and the @. At 15 and 20 the line turned into an even gray band.

Since Radius scales with the picture, blur first and crop after. On a crop 300 pixels tall, Radius 20 only reaches 17 pixels. That's a little more than Soft reaches on a full 1080-pixel screenshot, and the same 16-pixel line can still be made out.

Two white strips with the line alice.smith@gmail.com over the sky whale photo: the upper strip, under a Light blur box, still reads, and the lower one, under a Heavy box, is a gray smudge
Caption at Size 0.03 on a white plate, then Blur on a Rectangle mask: Light (Radius 2) above, Heavy (Radius 15) below

Settings for one line of a screenshot, a spotlight and a face

One line: set Height to about 4% for a line of interface text on a 1080-pixel-tall screenshot, Width wide enough to cover the words, Feather to 1% and Radius to 20. Keep the words at least a Feather's width inside the box. The blur fades back into the sharp picture across that band, and the very edge of the box is still half original. At 1% of a 1080-pixel picture, that band is 11 pixels.

Spotlight: tick Invert and bring Radius down to 5, the Soft preset. Everything outside the box goes soft, so the eye lands on one part of the screenshot and the rest is still recognizable.

Face or round badge: set Mask to Ellipse and Feather to 5% for an oval with soft edges.

Can blurred text be read again?

Sometimes, yes. A blur is a fixed weighted average of nearby pixels, so the same text in the same font always blurs the same way. If someone can guess the font and size, they can render candidate words, blur them identically and compare. A 2016 study by Hill, Zhou, Saul and Shacham got text back from blurred and pixelated documents like this, and concluded neither is fit for redaction.

It works because the blurred pixels still differ a little from one letter to the next. In the test above, changing a single letter, smith to smyth, still changed the Radius 20 blur in 6,781 pixels by up to 5 of 255 levels. One changed digit in a card number changed 2,222 pixels. Pixelating the address in 24-pixel blocks left 1,152 pixels different, and a PNG keeps every one of those levels.

A solid box keeps nothing of what's under it. For passwords, keys and account numbers, paint one over them in any image editor or crop them out, and use the blur for names, faces and clutter.

Saving a blurred screenshot

Save it as PNG. That keeps the rest of the screenshot's small type crisp where JPG would smudge it, and the file only holds the blurred pixels, with no hidden layer underneath.

On a video the box stays in one place. That's fine for a fixed caption or a corner, but it won't follow something moving across the frame. To put a label over the patch, open the saved PNG in Add Text to Image.

Further reading